Syncretism Matthew Baerman Surrey Morphology Group, University of Surrey
[email protected] 1 INTRODUCTION. If inflectional morphology were perfectly straightforward, a listing of the inflected forms of a word for a given language would at the same time give us a list of the morphosyntactic values relevant for that language. For example in Yir-Yoront (1), a PamaNyungan language from the Cape York Peninsula, Australia, the distinct forms displayed by such words as ‘foot’ or ‘leg’ justify distinguishing an absolutive, ergative and dative case. (1)
Case forms in Yir-Yoront (Alpher 1991: 211, 345, 487, 503) ‘foot’ ‘leg’ ‘arm’ ‘armpit’ ABSOLUTIVE thaml kumn puth ngamrr ERGATIVE thamarr kumalh putha ngumurr DATIVE thamarriy kuman putha ngumurr On the other hand, words such as ‘arm’ and ‘armpit’ fail to make the distinction between ergative and dative -- these words then show syncretism of case. This apparent merger of morphosyntactic values in a single form can be interpreted in two ways. On the one hand, it might simply be an indication that morphological structure may diverge from functional structure. On the other hand, there has long been an intuition that this identity reflects some fundamental affinity at the level of meaning or function (Plank 1991). In formal terms, we might view it as a window into the otherwise covert structure of morphosyntactic feature systems, i.e. revealing the building blocks of functional or semantic structure. It is safe to say that, for linguistic theory as a whole, it is the latter interpretation which has attracted the most attention, particularly since Jakobson (1936) and Hjelmslev (1935-37). On this approach, distinct morphosyntactic values may collapse if they overlap in meaning or function, and it is this shared element which is expressed by the single form. For example, in the Dagestanian language Xinalug, masculine and feminine gender are distinct in the singular, but typically combined in the plural, thus xækindædumæ ‘he’ll laugh’, xæskinkkudæmæ ‘she’ll laugh’, but xæpkindæmæ ‘they’ll laugh (masculine or feminine)’ (Dešeriev 1959:56). This syncretism is readily analysed (‘masculine’ + ‘feminine’ = ‘human’), and represents a cross-linguistically common pattern, so it is probably not a language-specific morphological quirk. Comparing this sort of data from various languages, one can start to build up a picture of shared aspects of feature structure (see e.g. Aikhenvald and Dixon 1998, Cysouw 2003, Baerman, Brown and Corbett 2005 and Iggesen 2005). But not all syncretism lends itself to a straightforward interpretation, and languages abound in patterns which are obscure in some way or other. It is the purpose of the present article to present a typology of patterns which are attested, and consider what implications these potentially have for models of morphology and morphosyntactic feature structure. I start with simple patterns that are relatively easily described, and move on to more complex patterns that require more morphological machinery. Within individual theoretical models some types may assume greater importance than others; likewise, some may turn out to be more common than others. But a satisfactory theory should be able to accommodate all types, or convincingly demonstrate that it does not need to. Of course, syncretism is simply one aspect of the larger question of how morphological form relates to function; an understanding of this smaller issue is crucial for understanding the larger one.
The preparation of this article was supported by the Arts and Humanities Research Council (UK) under grant number AH/D001579/1. Their support is gratefully acknowledged.
2 FLAT FEATURE STRUCTURE. The simplest type of syncretism can be described by assuming a flat (unstructured) feature system and invoking underspecification. Take the present tense of an English verb, which contrasts a 3SG form to one which collapses 1SG, 2SG and all plural values, e.g. sings versus sing. Compare this to the preterite form sang, which does not distinguish person-number at all. Since sang and sing are morphologically similar, in as much as they both lack an overt person-number ending, it is tempting to give them the same morphological analysis, and say that sing does not so much collapse the values of 1SG/2SG/1PL/2PL/3PL as simply fail to mark person-number at all. Sings is 3SG, while sing is simply a morphological default, used in those contexts where the more specific form is not already called for. This general principle is familiar under various names, such as the Elsewhere Condition (Kiparsky 1973), Paninian determinism (Stump 2001), blocking (Aronoff 1976) or the Subset Princple (Halle 1997). While the idea that there is such a thing as a morphological default is itself relatively neutral, it may serve as the basis for two further assumptions that begin to have real consequences for what can and cannot be described The first assumption is that morphosyntactic markedness correlates with semantic markedness (see e.g. Carstairs-McCarthy 1998, Harley and Ritter 2002, Burzio 2005). For any feature there is a default value, which is the semantically most basic one, e.g. singular for number or third for person. In a maximally economical system, these default values need not be overtly represented in a morphological description, since they are inserted by a redundancy rule. By this logic, the description of a six-member verbal paradigm, such as that of the Spanish present (2) need contain overt reference only to 1st person, 2nd person and plural, while the values ‘singular’ and ‘3rd person’ are supplied by default. (2)
Spanish present tense (vivér ‘live’) underspecified representation 1 1SG vivo 2SG vives 2 3SG vive ----1PL vivimos 1PL 2PL vivís 2PL viven PL 3PL
This mode of representation has underspecification already built into it. Syncretism naturally follows if we underspecify for further values. For example, in the imperfect, 1SG and 3SG are regularly syncretic in Spanish. If we say that 1SG is underspecified, then it follows automatically that it is syncretic with the 3SG. (3)
Spanish imperfect tense (vivér ‘live’) underspecified representation 1SG vivía ----2SG vivías 2 ----3SG vivía 1PL vivíamos 1PL 2PL vivíais 2PL vivían PL 3PL
For such an approach to have explanatory value, there would need to be some more or less generally agreed upon model of the markedness relations between morphosyntactic
2
values which we would expect the data to conform to. However, this would not seem to be a promising enterprise, as there is no demonstrable consistency across different languages. Consider the verb paradigms from Tol, an unclassified language of Honduras, given in (4) and (5). Tol has two main conjugation classes, labelled ‘I’ and ‘II’. The present tense paradigm of both classes has six forms, distinguishing 1st, 2nd and 3rd person in both singular and plural (likewise the past, which is distinguished from the present only by a prefix). The future tense paradigm for both classes is reduced: class I has four distinct forms and class II has five. The forms relevant for the present discussion are those ending in -kh in class I and those in -m in class II. (4)
Tol, class I verb, vocalic stem ‘put’ (Dennis 1992: 41, 51) PRESENT SINGULAR
1 2 3 (5) 1 2 3
ʔyonsó ʔyonson ʔyonso
FUTURE
PLURAL h
ʔyonsok ʔonsó haʔonsoph
SINGULAR
PLURAL
ka moʔonsó ka moʔonson ka moʔonsos
ka moʔonsokh ka moʔonsó ka moʔonsokh
Tol, class II verb ‘jump’ (Dennis 1992: 77, 79) PRESENT SINGULAR PLURAL
SINGULAR
tolos tolom tolo
ka ntolos ka ntolom ka ntolom
tyolokekh toloké teloŋ
FUTURE PLURAL
ka ntolokekh ka ntoloké ka ntoloŋ
In class I, -kh is 1PL in the present but 1PL/3PL in the future. In class II, -m is 2SG in the present but 2SG/3SG in the future. On analogy with the analysis offered above for Spanish, this could be attributed to underspecification: -kh is simply the unmarked plural ending for both classes, and -m is the unmarked singular ending for class II. Their actual distribution is constrained by the presence of more specific endings: class I ‘3PL PRESENT’ -ph blocks the appearance of the underpecified ‘PL’ -kh, and the class II ‘3 PRESENT’ -Ø blocks the appearance of the completely underspecified -m. But note that this analysis requires overt reference to 3rd person, which means that the universal assumption that 3rd person is unmarked will not hold. Worse still, the apparent markedness values vary across the classes: in class I, it is the 1st person which is unmarked, while in class II, it is 2nd person.
3
The second commonly-made assumption involving markedness is that morphosemantic markedness (in the sense of being overtly specified for a morphosyntactic value) correlates with morphological markedness (in the sense of bearing some overt bit of morphology; see Croft 2003). From this we should expect to find overtly marked forms which are non-syncretic, and forms without overt morphology which are syncretic, following the elsewhere condition. The English forms cited above (sing versus sing-s) are an obvious example. But it is easy enough to find paradigms where a zero realization is found, but it is not associated with the syncretic values. For example, Siuslaw (6), an unclassified language formerly spoken in Oregon, has enclitics which mark subject person-number, whereby the 2SG and 3PL are identical. The 3rd person marker is zero, but it is not syncretic. (6) 1INCL 1 2 3
Siuslaw (Frachtenberg 1922: 468) SINGULAR
DUAL
PLURAL
--------n -nx Ø
-ns -auxûn, -axûn, -ts -aux
-nł -nxan -tcî -nx
Thus, the association of zero morphology with syncretism is not a particularly robust generalization. 3 COMPLEX FEATURE STRUCTURE. If feature structure is assumed to be flat, there can only be one default value for any feature. Hence, if syncretism is seen as parasitic on feature structure, there can be at most one instance of syncretism for a feature within a paradigm.1 Examples such as the Sorbian noun in (7) are then problematic. In Sorbian, a West Slavic language of Germany, there is case syncretism in all three numbers. The singular and plural each have only a single syncretic form, which is unproblematic, but the dual has two: nominative/accusative and dative/locative/instrumental. (7)
A-stem noun (‘woman’) in Upper Sorbian (Fasske 1981: 493, 496, 498)
NOMINATIVE ACCUSATIVE GENITIVE LOCATIVE DATIVE INSTRUMENTAL
SINGULAR
DUAL
PLURAL
žona žonu žony žonje žonje žonu
žonje žonje žonow žonomaj žonomaj žonomaj
žony žony žonow žonach žonam žonami
If syncretism represents a default feature value, then the feature must have two defaults which are distinguished from each other in some way; that is, the feature has some internal structure. For example, to account for the Sorbian dual, we can set up an intermediate default shared by the dative, locative and instrumental, as in table 1, to which we might impute the value ‘oblique’. On this model, the nominative and accusative forms of the dual are fully unspecified and default to CASE, the genitive form is fully specified, while the dative, locative and instrumental forms are underspecified as OBLIQUE.
1
By ‘paradigm’ I mean the variants along one parameter, with all other parameters remaining constant. Thus in (7) one can speak of a distinct case paradigm for each number, and a distinct number paradigm for each case.
4
CASE
OBLIQUE
NOM
ACC
GEN
DAT
LOC
INS
Table 1. A possible model of case in Sorbian It is probably fair to say that much of the interest in syncretism shown by contemporary linguistics is stimulated by this possibility of applying syncretism as a diagnostic of otherwise covert aspects of feature structure (see e.g. Calabrese 1998, Cysouw 2003, Müller 2004, Wiese 2004). Various ways of construing this structure have been proposed -- trees, networks, binary feature values and the like, each with their own inherent constraints (see Johnston 1997 and Harbour 2007 for discussion of various systems and their properties). 4 STIPULATION. If syncretism is assumed to reflect a covert feature value, then it is reasonable to expect that the value thus revealed is a semantically plausible one. While there is no universal yardstick for determining what constitutes a plausible feature value, there are examples of syncretism where the syncretized values clearly do not form a natural class in any readily comprehensible terms. For example, verbs in Dhaasanac (8), a Cushitic language spoken in Ethiopia, display a maximum of two person-number-gender forms, one for 1PL (exclusive), 2nd person, and 3rd person feminine, and one for 1SG , 1st person inclusive, 3PL feminine and 3rd person masculine. (The alternation between the two forms is morphologically heterogeneous -- suffixation, prefixation, ablaut, consonant alternation -- the example illustrates just one of the many types.) (8)
Dhaasanac (Tosco 2001: 112, 456)
1INCLUSIVE 1 2 3 FEMININE 3 MASCULINE
SINGULAR
PLURAL
-----seð sieti sieti seð
seð sieti sieti seð seð
Though one of these two patterns might be treated as an elsewhere form, the other must still be accounted for, and no model of feature structure yet proposed will combine such values. Since the pattern is clearly systematic, it must be encoded somewhere in the grammar. If it is not due to feature structure, then it must simply be stipulated in the morphology. Of course, if one were committed to the dictum that ‘form ever follows function, and this is the law’ (Sullivan 1896) then one could presumably construct a feature structure which would yield the desired results. At a certain stage, though (and Dhaasanac surely is beyond that stage), such an exercise would become unedifying. Indeed, there is overwhelming evidence for morphological units which operate at cross-purposes to syntax and semantics (Maiden 1992, Aronoff 1994, Blevins 2006).
5
One way of representing this is through simple disjunction, i.e. the syncretism of features ‘x’ and ‘y’ can be represented as {x or y} (as in Dalrymple and Kaplan 2000). Alternatively, one might imagine morphological structure as a system parallel to morphosyntactic structure. One would simply need to substitute the idea of morphological features (Spencer and Sadler 2001, Corbett and Baerman 2007) for that of morphosyntactic features. For example, we might propose that the Dhaasanac verbal paradigm has the shape in table 2, where the 1PL, 2nd person and 3SG feminine forms share the morphological feature ‘B’, while the other forms share the (default) morphological feature ‘A’ (see Baerman, Brown and Corbett 2005: 185). A
B
FORM 1 (1SG)
FORM 2 (3SG M)
FORM 3 (1INCL)
FORM 4 (3PL)
FORM 5 (1PL)
FORM 6
(2)
FORM 7 (3SG F)
Table 2. A possible model of the morphological structure of person marking in Dhaasanac These features have no inherent meaning, they simply codify the organizing principles of the paradigm of forms. Such an approach presupposes the existence of morphology as a distinct component of grammar, but does not presuppose any specifically morphological devices; everything is achieved by underspecification. However, there are limits to what underspecification can express. An especially tricky example comes from second declension nouns in Latin (9). This declension class has both masculines and neuters. The neuters, as in other Indo-European languages, show nominative/accusative syncretism. In the singular this is realized as -um, the same ending that serves for the distinct accusative of the masculines. So far, this is a perfectly banal instance of syncretism, which could be analyzed by any of the devices described above. For example, we might say that the form in -um can serve for both nominative and accusative, but in the paradigm where a more specific nominative form in -us is available, the form in -um is restricted to the accusative. Things get more interesting when we consider the behaviour of a tiny class of exceptional neuters in -us, such as pelagus ‘sea’.2 As expected for neuters, nominative and accusative are syncretic, but the form ends in -us rather than -um.
2
It consists of the native words vulgus ‘crowd’ and vīrus ‘poison’ and the word ‘sea’, which is a borrowing from Greek. Other Greek borrowings in -us may occasionally receive the same treatment, such as cētus ‘large sea animal, whale’ and chaus ‘chaos’; otherwise, they take a distinct accusative in -um (Neue and Wagener 1902: 502-504).
6
(9)
Latin second conjugation (singular forms) masculine normal neuter ‘slave’ ‘war’ NOMINATIVE serv-us bell-um ACCUSATIVE serv-um bell-um GENITIVE serv-ī bell-ī DATIVE serv-ō bell-ō
neuter in -us ‘sea’ pelag-us pelag-us pelag-ī pelag-ō
The problem is this: a simple underspecification account requires that forms in -us be overtly specified as nominative, but this is belied by the behaviour of nouns such as pelagus, which shows that forms in -us may be syncretic. If both -um and -us are nominative/accusative, how can we account for their distribution in the paradigm of servus? The only systematic way to represent this pattern across the three sub-classes is to separate the forms from their morphosyntactic distribution. For neuters, nominative and accusative are always syncretic, but the form that this takes varies. For most items, it takes the form otherwise displayed by the accusative, but for the small class of nouns like pelagus, it takes the form of the nominative. This redistribution of forms has been called a ‘rule of referral’ (Zwicky 1985, Corbett and Fraser 1993, Stump 2001), ‘takeover’ (Carstairs 1984) or a particular type of readjustment rule (Halle 1992). What distinguishes this from the underspecification approaches described above is that this relationship is at cross-purposes with the rest of morphological structure, and so must simply be stipulated. The status of such rules is controversial, viewed by some as a superfluous morphological device (Wunderlich 2004). 5 DIACHRONY. The abstract question of how best to analyze syncretic patterns becomes more concrete when we look not at static examples, but at dynamic systems, in particular, instances of diachronic change. In some cases these provide evidence of an actual collapse of feature values at the morphosyntactic level, while others involve a merger of only of the forms themselves. Feature-based syncretism is most clearly represented where one finds the apparently spontaneous reorganization of the system of morphosyntactic values. Consider the development of the dual and plural from Common Slavic to Slovenian. The changes can be attributed to regular sound change (denasalization of ẽ, loss of word-final ĭ and ŭ), except for the genitive and locative dual. In place of the expected ending -u, the corresponding forms of the plural are found. (10)
Development of the genitive and locative dual in Slovenian Common Slavic Slovenian
NOMINATIVE ACCUSATIVE GENITIVE LOCATIVE DATIVE INSTRUMENTAL
DUAL
PLURAL
DUAL
PLURAL
duši duši dušu dušu dušama dušama
dušẽ dušẽ dušĭ dušaxŭ dušamŭ dušami
duši duši duš dušah dušama dušama
duše duše duš dušah dušam dušami
This change should be seen in the wider context of the loss of the dual as a number value, familiar from other Indo-European languages. That is, it is probably a reflection of the incipient consolidation of the values dual and plural at the morphosyntactic level. Further discussion of feature-based syncretism can be found in Luraghi (2000).
7
Form-based syncretism is most clearly seen where a regular sound change produces homophony for some forms which is then extended to others. One fairly clear example is represented by the 1SG in Livonian, a Finnic language formerly spoken in Latvia. In an earlier period, the 1SG ending was -n in both present and past, a state of affairs still found in Estonian. With the regular loss of final -n the 1SG fell together with the endingless 3SG in the past. In the present, where the 3SG ending is -b, no homophony should have resulted. But what we find is that 3SG -b has been extended to the 1SG, i.e. the superficial pattern of syncretism of the past has been extended to the present. (11) 1SG 2SG 3SG 1PL 2PL 3PL
Livonian (Kettunen 1938: LX) compared to Estonian (verb ‘kill’) Livonian Estonian PRESENT
PAST
PRESENT
PAST
tapab tapad tapab tapam tapat tapabəd
tapiz tapist tapiz tapizm(ə) tapist(ə) tapist(ə)
tapan tapad tabab tapame tapate tapavad
tapsin tapsid tapis tapsime tapsite tapsid
This suggests that the patterns that result from phonological accidents may be construed as systematic, and so enter into the morphological system. The syncretism of second and third person in the plural found in the preterite, likewise an example of regular sound change, seems also to have been partially extended to the present, being restricted to the negative paradigm (Kettunen 1938: LXI).3 See Hansson (to appear) for a similar but more complex change in the case system of Saami. More often than not, though, it is hard to tease apart morphosyntactic and morphological change. Bazell observed (1960: 1) that ‘simultaneous morphemic and phonological similarity’ tend to attract each other. For example, the weak masculine noun paradigm in Old English (Mercian and West Saxon, to be precise) shows massive syncretism: a single ending -an is shared by the accusative, genitive and dative singular and the nominative and accusative plural (12a). This is not a result of regular sound change, which should have produced a more differentiated paradigm (12b).4
3
This is formed with a proclitic negative auxiliary. In the singular, the main verb is uninflected, while in the plural, the 3PL has the same form as the 2PL. The auxiliary itself shows number syncretism, with the plural forms inflected as singulars (Kettunen 1938: LXI): 1SG 2SG 3SG 1PL 2PL 3PL
‘don’t read’ äp=tapa ät=tapa äp=tapa äp=tapam ät=tapat äp=tapat
(Note that the final consonant of the auxiliary is devoiced when followed by a voiceless consonant.) Compare this with the behavior of verbal forms, e.g. helpan ‘help.INF’ versus hulpon ‘helped.PL’ (Sievers 1921: 199). 4
8
(12)
Endings of weak masculine nouns in Old English (Bazell 1960: 3) a. attested b. expected
NOMINATIVE ACCUSATIVE GENITIVE DATIVE
SINGULAR
PLURAL
SINGULAR
PLURAL
-a -an -an -an
-an -an -ena -um
-a -on -an -an
-an -on -ena -um
It does not seem possible to ascribe this change to one single component; rather, it results from a conspiracy of functional and formal analogies. 6 CONCLUSION. In the preceding sections I have tried to show to what extent syncretism can be derived from morphosyntactic structure on the one hand, and how much needs to be attributed to purely morphological structure on the other. It is not a crystal-clear picture which results, since the bulk of examples one can find will easily lend themselves to alternative analyses. In the light of this, there are the three approaches which I believe it is possible to take: 1. Ascribe syncretism wholly to morphosyntactic features. A theory of syncretism which derives it from morphosyntactic features is more restrictive, and so is to be preferred, all else being equal (Noyer 1998, Bobaljik 2002). However, no model of feature structure has yet been advanced that will account for the attested patterns of syncretism, and patterns such as that in (9) above suggest that none ever will. Therefore, if this approach is to be maintained, some principled treatment of the exceptions must be found.5 2. Ascribe syncretism wholly to morphology. On the assumption that morphological structure is language-specific, this approach will allow any pattern to be described. However, it leaves open the question as to why some syncretic patterns are commonly more attested crosslinguistically than others. Therefore, such an approach has to clarify exactly how morphology is linked to morphosyntactic structure, and what (if anything) governs their correspondence to each other. 3. Admit indeterminacy. There is good evidence both for a morphosyntactic and a morphological approach to syncretism, but the bulk of examples are indeterminate, and admit of either interpretation. Rather than treat this as an embarrassment, one might take the indeterminacy as a characteristic of inflectional paradigms. Indeed, developments such as that found in Old English (12 above) suggest that this is the case. The third approach, I believe, holds the most promise. Since the analysis of static examples will do little to resolve the issue of what lies behind the ambiguity, effective research in this direction should focus on dynamic systems, that is, across variants. The most time-tested tool for looking at this is historical linguistics, as suggested in §5, but other disciplines which treat variation might be brought to bear too, such as dialectology (Aalberse, to appear), acquisition studies or psycholinguistics.
5
Understandably, restrictive theories of syncretism make claims about systematic syncretism, but not about accidental homophony. While accidental homophony surely exists, there are no ready diagnostics for identifying it. In particular, the diachronic continuum discussed above with respect to Livonian (figure (11)) shows that the accidental can be reinterpreted as systematic.
9
REFERENCES Aalberse, Suzanne (to appear). The typology of syncretisms and the status of feature structure: verbal paradigms across 355 dialects. Morphology. Aikhenvald, Alexandra Y. and Robert M. W. Dixon. 1998. Dependencies between grammatical systems. Language 74. 56-80. Alpher, Barry. 1991. Yir-Yoront lexicon: sketch and dictionary of an Australian language. Berlin: Mouton de Gruyter. Aronoff, Mark. 1976. Word formation in generative grammar. Cambridge, MA: MIT Press. Aronoff, Mark. 1994 Morphology by itself. Cambridge: MIT Press. Bazell, C. E. 1960. A question of syncretism and analogy. Transactions of the Philological Society. 1-12. Baerman, Matthew, Brown, Dunstan, and Corbett Greville G. 2005. The syntax-morphology interface: a study of syncretism (Cambridge Studies in Linguistics 109). Cambridge: Cambridge University Press. Blevins, James P. 2006. Word-based morphology. Journal of Linguistics 42.3, 531–573. Bobaljik, Jonathan. 2002. Syncretism without paradigms: remarks on Williams 1981, 1994. Yearbook of Morphology 2001, ed. by Geert Booij and Jaap van Marle, 53-86. Dordrecht: Kluwer. Burzio, Luigi. 2005. Sources of paradigm uniformity. Paradigms in phonological theory, ed. by Laura J. Downing, T. A. Hall and Renate Raffelsiefen, 65-106. Oxford: Oxford University Press. Calabrese, Andrea. 1998. Some remarks on the Latin case system and its development in Romance. Theoretical analyses on Romance languages, ed. by José Lema and Esthela Treviño, 71-126. Amsterdam/Philadelphia: John Benjamins. Carstairs, Andrew. 1987. Allomorphy in inflection. London: Croom Helm. Carstairs-McCarthy, Andrew. 1998. How lexical semantics constrains inflectional allomorphy. Yearbook of Morphology 1997, ed. by Geert Booij and Jaap van Marle, 1-24. Dordrecht: Kluwer.. Corbett, Greville G., and Baerman, Matthew. 2007. Prolegomena to a typology of morphological features. Morphology. Corbett, Greville G. and Norman M. Fraser. 1993. Network morphology: a DATR account of Russian nominal inflection. Journal of Linguistics 29. 113-42. Croft, William. 2003. Typology and universals. Cambridge: Cambridge University Press.
10
Cysouw, Michael. 2003. The paradigmatic structure of person marking. Oxford: Oxford University Press Dennis, Ronald K. 1992. La conjugación del verbo en idioma tol (jicaque). Tegucigalpa: Instituto Hondureño de Antropología e Historia and Instituto Lingüístico de Verano. Dešeriev, Ju. D. 1959. Grammatika xinalugskogo jazyka. Moscow: Akademija Nauk SSSR. Fasske, Helmut. 1981. Grammatik der obersorbischen Schriftsprache der Gegenwart: Morphologie. Bautzen: Domowina Verlag. Frachtenberg, Leo J. 1922. Siuslaw. Handbook of American Indian Languages 2, ed. by Franz Boas, 431-629. Washington: Government Printing Office, Washington. Halle, Morris. 1997. Distributed morphology: Impoverishment and fission. MIT Working Papers in Linguistics 30: Papers at the Interface, ed. Benjamin Bruening, Yoonjung Kang and Martha McGinnis, 425-449. Cambridge: MIT. Hansson, Gunnar Ó. (To appear). Productive syncretism in Saami inflectional morphology. Saami Linguistics, ed. by Ida Toivonen, Diane Nelson and William Palmer. Amsterdam/Philadelphia: John Benjamins. Harley, Heidi and Elizabeth Ritter. 2002. Person and number in pronouns: a feature geometric analysis. Language 78.482 526. Harbour, Daniel. 2007. Morphosemantic number: from Kiowa noun classes to UG Number features. Dordrecht: Springer. Hjelmslev, Louis. 1935-37. La catégorie des cas: étude de grammaire générale. Aarhus: Universitetsforlaget,. Iggesen, Oliver. 2005. Asymmetrical case marking. World atlas of language structures, ed. by Martin Haspelmath, Matthew Dryer, David Gil and Bernard Comrie, 206-209. Oxford: Oxford University Press. Jakobson, Roman O. 1936 (1971). Beitrag zur allgemeinen Kasuslehre: Gesamtbedeutung der russischen Kasus. Reprinted in: Selected writings (vol. II: Word and language), 23-71. The Hague: Mouton . Johnston, Jason. 1997. Systematic homonymy and the structure of morphological categories: some lessons from paradigm geometry. PhD Thesis, University of Sydney. [available online at http://ses.library.usyd.edu.au/handle/2123/396] Dalrymple, Mary and Ronald M. Kaplan. 2000. Feature indeterminacy and feature resolution. Language 76/4. 759-798. Halle, Morris. 1992. The Latvian declension. Yearbook of Morphology 1991, ed. by Geert Booij and Jaap van Marle, 33-48. Dordrecht: Kluwer.
11
Kettunen, Lauri. 1938. Livisches Wörterbuch mit grammatischer Einleitung. Helsinki: Suomalais Ugrilainen Seura. Kiparsky, Paul. 1973. ‘Elsewhere’ in phonology. A Festschrift for Morris Halle, ed. by Stephen R. Anderson and Paul Kiparsky, 93-106. New York: Holt, Reinhart, and Winston. Luraghi, Silvia. 2000. Synkretismus. Morphologie: ein Handbuch zur Flexion und Wortbildung, ed. by Geert Booij, Christian Lehmann and Joachim Mugdan, 638-647. Berlin: Mouton de Gruyter. Maiden, Martin 1992. Irregularity as a determinant of morphological change. Journal of Linguistics 28/2. 285-312. Müller, Gereon. 2004. A Distributed Morphology approach to syncretism in Russian noun inflection. Proceedings of Formal Approaches to Slavic Linguistics 12, ed. by In: Olga Arnaudova, Wayles Browne, Maria Luisa Rivero and Dejan Stojanović, 353-373.. Ann Arbor: Michigan Slavic Publications. Neue, Friedrich and C. Wagener. 1902. Formenlehre der lateinischen Sprache (vol. 1: Das Substantivum). Leipzig: O. R. Reisland. Noyer, Rolf. 1998. Impoverishment theory and morphosyntactic markedness. Morphology and its relation to phonology and syntax, ed. by Steven G. Lapointe, Diane K. Brentari, and Patrick M. Farrell, 264-285. Stanford: CSLI Publications. Plank, Frans. 1991. Rasmus Rask's dilemma. Paradigms: the economy of inflection, ed. by Frans Plank, 161-196. Berlin: Mouton de Gruyter. Sadler, Louisa and Andrew Spencer. 2001. Syntax as an exponent of morphological features. Yearbook of Morphology 2000, ed. by Geert Booij and Jaap van Marle, 71-96. Dordrecht: Kluwer. Sievers, Eduard. 1921. Angelsächsische Grammatik. Halle: Max Niemeyer. Stump, Gregory T. 2001. Inflectional morphology. Cambridge: Cambridge University Press. Sullivan, Lous H. 1896 (1980). The tall office building artistically considered. Lippincott's Magazine, March. Reprinted in: Kindergarten chats and other writings, 202-213. New York: Dover Publications. Tosco, Mauro. 2001. The Dhaasanac language. Cologne: Rüdiger Köppe. Wiese, Bernd. 2004. Categories and paradigms: on underspecification in Russian declension. Explorations in nominal Inflection, ed. by Gereon Müller, Lutz Gunkel, & Gisela Zifonun, 321-372. Berlin: Mouton de Gruyter, Berlin. Wunderlich, Dieter. 2004. Is there any need for the concept of directional syncretism?. Explorations in nominal Inflection, ed. by Gereon Müller, Lutz Gunkel, & Gisela Zifonun, 373-395. Berlin: Mouton de Gruyter.
12
Zwicky, Arnold. 1985. How to describe inflection. Proceedings of the eleventh annual meeting of the Berkeley Linguistics Society. 372-386.
13